Assignment-8: Second component
Effect of Education in Determining Race, Gender Income Inequality across States in the USA
Introduction
Racial, gender wage gaps persist in U.S. Research revealed that women of all major racial and ethnic groups earn less than men of the same group, and also earn less than White men. Education is one of the most influential determinants of person’s wage, and thus might be a contributing factor to the racial wage gap. The effect of education affects wage differently among various racial groups. Education affects wages since it allows access to occupations of higher status which offers higher earnings. The black-white wage gap has been decreasing due to the decrease in education gap. The distance from home and job locations was also found to affects the ability of minorities to find profitable work. In addition, the redistribution of manufacturing jobs location also led wage gap between blacks and whites depending on whether they live in cities or not. So, The present study was designed to answer the following questions.
Research questions: 1. To what extent education mediates the income inequality of Black and White combined with gender. 2. To what extent effect of education in income inequality of Black and White combined with gender vary across the States.
Method
Sampling method:
The sample was drawn from the population the American Community Survey data 2016-2012. The data was extracted from ipumps.org. The sample included employed persons between the age range of 18-66 years.People only from Black and White race groups were included in the sample. The sample was from 46 different States of USA.
Defining variables
Dependent variable
The income was the dependent variable which indicates each respondent’s total pre-tax personal income or losses from all sources for the previous year. The income was converted to thousand for analysis.
Independent variables(Level-1)
Four independent variables were used in this study. The variable “gender” reports whether the person was male or female. The variable “education” indicates respondents’ educational attainment, as measured by the highest year of school or degree completed.The variable “race” indicated the person’s major race groups: White or Black. The variable “UHRSWORK” is a 2-digit numeric variable that reports the number of hours per week that the respondent usually worked if the person worked during the previous year.
Independent variable(Level-2)
Although the sample included people from 46 different States, only 42 state used for analyzing the results. 4 states were reported with either 0 or very few Black people and were excluded to avoid convergence errors.
Control variable
Person’s age was included in the regression models to get the best fitting model for analysis.
Data analysis
Procedure of data analysis
Multilevel logistic regression analysis was conducted by using 5 different models to get the best-fitted model. AIC measures were used for identifying the best fitting models. The best-fitting model was simulated for obtaining random variance and probability interval.
library(tidyverse)
income<-read_csv("income.csv")
head(income)## # A tibble: 6 x 35
## YEAR DATANUM SERIAL HHWT REGION STATEFIP GQ PERNUM PERWT FAMSIZE
## <int> <int> <int> <int> <int> <int> <int> <int> <int> <int>
## 1 2012 1 135 28 32 1 1 2 77 8
## 2 2012 1 184 45 32 1 1 1 45 4
## 3 2012 1 338 108 32 1 1 1 108 6
## 4 2012 1 338 108 32 1 1 2 174 6
## 5 2012 1 338 108 32 1 1 4 417 6
## 6 2012 1 338 108 32 1 1 5 272 6
## # ... with 25 more variables: NCHILD <int>, NCHLT5 <int>, ELDCH <int>,
## # YNGCH <int>, SEX <int>, AGE <int>, MARST <int>, RACE <int>,
## # RACED <int>, CITIZEN <int>, RACASIAN <int>, RACBLK <int>,
## # RACWHT <int>, HCOVANY <int>, EDUC <int>, EDUCD <int>, EMPSTAT <int>,
## # EMPSTATD <int>, UHRSWORK <int>, INCTOT <int>, FTOTINC <int>,
## # INCEARN <int>, POVERTY <int>, MIGRATE1 <int>, MIGRATE1D <int>sub_income<-income[c(6,15, 16,18,25, 27, 29, 30, 31)]sub_income<-sub_income %>%
mutate(race = sjmisc::rec(RACE, rec = "1=0; 2=1"))%>%
mutate( gender= sjmisc::rec(SEX, rec = "1=0; 2=1"))sub_income$gender<-factor(sub_income$gender, levels = c(0, 1),
labels = c("Male","Female"))
sub_income$race<-factor(sub_income$race, levels = c (0, 1),
labels = c("White", "Black"))library(dplyr)
sub_income1<-subset(sub_income, EMPSTAT==1)
head(sub_income1)## # A tibble: 6 x 11
## STATEFIP SEX AGE RACE EDUC EMPSTAT UHRSWORK INCTOT FTOTINC race
## <int> <int> <int> <int> <int> <int> <int> <int> <int> <fct>
## 1 1 2 27 1 6 1 40 4000 152000 White
## 2 1 2 35 1 6 1 40 45000 45000 White
## 3 1 1 31 1 6 1 40 33000 96000 White
## 4 1 1 26 1 6 1 40 63000 96000 White
## 5 1 2 31 1 11 1 40 44000 67000 White
## 6 1 2 36 2 7 1 14 18950 18950 Black
## # ... with 1 more variable: gender <fct>sub_income2<-sub_income1 %>% mutate(c_edu = EDUC- median(EDUC)) %>%
mutate(c_age = AGE- median(AGE))%>%
mutate(income=INCTOT/1000)%>%
mutate(c_hour=UHRSWORK- median(UHRSWORK))w=xtabs(~STATEFIP+race, sub_income2)
w[1:46, 1:2]## race
## STATEFIP White Black
## 1 466 94
## 2 91 7
## 4 2511 192
## 5 418 26
## 6 21381 892
## 8 1886 220
## 9 1308 381
## 10 200 82
## 11 244 113
## 12 11241 2856
## 13 2496 900
## 15 149 20
## 16 360 8
## 17 4512 327
## 18 763 101
## 19 412 56
## 20 593 66
## 21 518 89
## 22 489 89
## 23 142 23
## 24 1914 1370
## 25 2726 881
## 26 1631 131
## 27 670 333
## 28 189 22
## 29 638 134
## 30 76 1
## 31 307 51
## 32 1273 131
## 33 255 22
## 34 4636 1045
## 35 540 18
## 36 7201 3191
## 37 2315 424
## 38 69 16
## 39 1200 330
## 40 641 40
## 41 1135 74
## 42 1541 444
## 44 306 121
## 45 865 97
## 46 82 18
## 47 1032 175
## 48 14024 1206
## 49 702 50
## 50 92 7sub_income3<-filter(sub_income2, STATEFIP %in% c(1,4:15, 17:29, 29, 31:49 ) )Results
Descriptive Analysis
summary(sub_income3$AGE )## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 18.00 31.00 36.00 36.79 42.00 66.00summary(sub_income3$EDUC)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 1.000 6.000 7.000 7.342 10.000 11.000summary(sub_income3$UHRSWORK)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 1.00 40.00 40.00 40.21 45.00 99.00Table1<-table(sub_income3$gender, sub_income3$race)
barplot(Table1, beside=T, legend.text = c("Male", "Female"))
The descriptive analysis of data revealed that the age range of the sample was 18-66 years and the median age of the sample was 36 years. The range of educational attainment varies between nursary to 5+ years of college education andthe median educational attainment was 7 years which was equivalent to 1-year of college education. The median weekly work hours was 40. White males were highly representative in the sample.
dcoef <- sub_income3%>%
group_by(STATEFIP) %>%
do(mod = lm(income ~ race+gender+c_edu+c_age+race*c_edu+gender*c_edu+race*c_hour, data = .))
coef <- dcoef %>% do(data.frame(intc = coef(.$mod)[1]))
ggplot(coef, aes(x = intc)) + geom_histogram()
dcoef <- sub_income3 %>%
group_by(STATEFIP) %>%
do(mod = lm(income ~ race+gender+c_edu+c_age+race*c_edu+gender*c_edu+race*c_hour, data = .))
coef <- dcoef %>% do(data.frame(intc = coef(.$mod)[2]))
ggplot(coef, aes(x = intc)) + geom_histogram()
Complete pooling model
cpooling <- lm(income ~ race+gender+c_edu+c_age+race*c_edu+gender*c_edu+race*c_hour, data = sub_income3)
summary(cpooling)##
## Call:
## lm(formula = income ~ race + gender + c_edu + c_age + race *
## c_edu + gender * c_edu + race * c_hour, data = sub_income3)
##
## Residuals:
## Min 1Q Median 3Q Max
## -201.11 -27.32 -8.02 12.96 1160.97
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 59.01030 0.26038 226.63 <2e-16 ***
## raceBlack -15.69007 0.57032 -27.51 <2e-16 ***
## genderFemale -13.14729 0.41753 -31.49 <2e-16 ***
## c_edu 11.01098 0.08817 124.88 <2e-16 ***
## c_age 1.32460 0.02577 51.39 <2e-16 ***
## c_hour 1.67004 0.01881 88.78 <2e-16 ***
## raceBlack:c_edu -2.47257 0.23380 -10.58 <2e-16 ***
## genderFemale:c_edu -4.74301 0.14088 -33.67 <2e-16 ***
## raceBlack:c_hour -0.71381 0.04763 -14.99 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 62.77 on 112463 degrees of freedom
## Multiple R-squared: 0.2762, Adjusted R-squared: 0.2761
## F-statistic: 5364 on 8 and 112463 DF, p-value: < 2.2e-16Random intercept model
library(lme4)
library(merTools)
library(lmerTest)
library(Matrix)
m1 <- lme4::lmer(income~ 1 + (1|STATEFIP), data=sub_income3)
summary(m1)## Linear mixed model fit by REML ['lmerMod']
## Formula: income ~ 1 + (1 | STATEFIP)
## Data: sub_income3
##
## REML criterion at convergence: 1284747
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.2743 -0.4851 -0.2728 0.1001 16.5383
##
## Random effects:
## Groups Name Variance Std.Dev.
## STATEFIP (Intercept) 155.3 12.46
## Residual 5343.7 73.10
## Number of obs: 112472, groups: STATEFIP, 42
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 52.72 1.98 26.62m2 <- lme4::lmer(income~ 1+race+gender+c_edu+c_age+race*c_edu+gender*c_edu+race*c_hour + (1|STATEFIP), data=sub_income3)
summary(m2)## Linear mixed model fit by REML ['lmerMod']
## Formula:
## income ~ 1 + race + gender + c_edu + c_age + race * c_edu + gender *
## c_edu + race * c_hour + (1 | STATEFIP)
## Data: sub_income3
##
## REML criterion at convergence: 1249249
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.1525 -0.4366 -0.1249 0.2074 18.4470
##
## Random effects:
## Groups Name Variance Std.Dev.
## STATEFIP (Intercept) 46.56 6.824
## Residual 3898.01 62.434
## Number of obs: 112472, groups: STATEFIP, 42
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 56.43124 1.13681 49.64
## raceBlack -16.49694 0.58892 -28.01
## genderFemale -13.15517 0.41577 -31.64
## c_edu 10.91097 0.08902 122.57
## c_age 1.27465 0.02578 49.44
## c_hour 1.67264 0.01874 89.26
## raceBlack:c_edu -2.41826 0.23406 -10.33
## genderFemale:c_edu -4.77298 0.14017 -34.05
## raceBlack:c_hour -0.69849 0.04740 -14.74
##
## Correlation of Fixed Effects:
## (Intr) rcBlck gndrFm c_edu c_age c_hour rcBlck:c_d gndF:_
## raceBlack -0.063
## genderFemal -0.140 -0.077
## c_edu -0.001 0.045 -0.050
## c_age -0.023 -0.043 0.160 -0.186
## c_hour -0.049 -0.011 0.297 -0.108 -0.017
## racBlck:c_d 0.004 -0.374 0.054 -0.221 0.019 0.043
## gndrFml:c_d 0.002 0.057 -0.174 -0.540 -0.002 -0.008 -0.079
## rcBlck:c_hr 0.007 0.080 -0.037 0.031 -0.015 -0.368 -0.115 0.009m3 <- lme4::lmer(income~ 1+race+gender+c_edu+c_age+race*c_edu+gender*c_edu+gender*c_hour +race*c_hour+ (1|SEX)+(1|RACE)+(1|STATEFIP), data=sub_income3)
summary(m3)## Linear mixed model fit by REML ['lmerMod']
## Formula:
## income ~ 1 + race + gender + c_edu + c_age + race * c_edu + gender *
## c_edu + gender * c_hour + race * c_hour + (1 | SEX) + (1 |
## RACE) + (1 | STATEFIP)
## Data: sub_income3
##
## REML criterion at convergence: 1249200
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.2255 -0.4353 -0.1235 0.2063 18.4524
##
## Random effects:
## Groups Name Variance Std.Dev.
## STATEFIP (Intercept) 46.73 6.836
## RACE (Intercept) 14.18 3.765
## SEX (Intercept) 108.82 10.432
## Residual 3896.16 62.419
## Number of obs: 112472, groups: STATEFIP, 42; RACE, 2; SEX, 2
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 56.04213 11.14887 5.03
## raceBlack -16.32564 5.35746 -3.05
## genderFemale -13.46768 14.75858 -0.91
## c_edu 10.84741 0.08942 121.32
## c_age 1.27453 0.02578 49.45
## c_hour 1.78409 0.02407 74.13
## raceBlack:c_edu -2.40524 0.23401 -10.28
## genderFemale:c_edu -4.62633 0.14154 -32.69
## genderFemale:c_hour -0.25938 0.03516 -7.38
## raceBlack:c_hour -0.68869 0.04741 -14.53
##
## Correlation of Fixed Effects:
## (Intr) rcBlck gndrFm c_edu c_age c_hour rcBlck:c_d
## raceBlack -0.238
## genderFemal -0.662 0.000
## c_edu 0.000 0.005 -0.001
## c_age -0.002 -0.005 0.005 -0.185
## c_hour -0.007 0.002 0.005 -0.144 -0.014
## racBlck:c_d 0.000 -0.041 0.001 -0.220 0.019 0.038
## gndrFml:c_d 0.000 0.007 -0.005 -0.546 -0.002 0.082 -0.077
## gndrFml:c_h 0.005 -0.004 0.003 0.096 0.001 -0.628 -0.008
## rcBlck:c_hr 0.001 0.009 -0.001 0.028 -0.015 -0.269 -0.114
## gndrFml:c_d gndrFml:c_h
## raceBlack
## genderFemal
## c_edu
## c_age
## c_hour
## racBlck:c_d
## gndrFml:c_d
## gndrFml:c_h -0.140
## rcBlck:c_hr 0.012 -0.028Random slope model
m4 <- lme4::lmer(income~ race*gender*c_edu*c_hour+c_age+ (1+gender+c_edu+race|STATEFIP), data=sub_income3)
summary(m4)## Linear mixed model fit by REML ['lmerMod']
## Formula: income ~ race * gender * c_edu * c_hour + c_age + (1 + gender +
## c_edu + race | STATEFIP)
## Data: sub_income3
##
## REML criterion at convergence: 1244154
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -4.4843 -0.3976 -0.1068 0.1852 18.6124
##
## Random effects:
## Groups Name Variance Std.Dev. Corr
## STATEFIP (Intercept) 60.049 7.749
## genderFemale 26.989 5.195 -0.94
## c_edu 6.214 2.493 0.95 -0.93
## raceBlack 17.474 4.180 -0.99 0.91 -0.98
## Residual 3722.394 61.011
## Number of obs: 112472, groups: STATEFIP, 42
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 53.017570 1.276718 41.53
## raceBlack -16.915962 1.038343 -16.29
## genderFemale -13.635683 0.996838 -13.68
## c_edu 8.473796 0.410308 20.65
## c_hour 1.641740 0.024581 66.79
## c_age 1.260268 0.025204 50.00
## raceBlack:genderFemale 12.856293 1.188031 10.82
## raceBlack:c_edu -3.222162 0.318543 -10.12
## genderFemale:c_edu -2.389654 0.157493 -15.17
## raceBlack:c_hour -0.786365 0.071366 -11.02
## genderFemale:c_hour -0.461917 0.038686 -11.94
## c_edu:c_hour 0.440357 0.008398 52.43
## raceBlack:genderFemale:c_edu 1.350413 0.470648 2.87
## raceBlack:genderFemale:c_hour 0.418472 0.104123 4.02
## raceBlack:c_edu:c_hour -0.160074 0.028130 -5.69
## genderFemale:c_edu:c_hour -0.135061 0.013366 -10.11
## raceBlack:genderFemale:c_edu:c_hour 0.104439 0.042476 2.46AIC(cpooling, m1, m2,m3, m4)## df AIC
## cpooling 10 1250329
## m1 3 1284753
## m2 11 1249271
## m3 14 1249227
## m4 28 1244210library(texreg)
htmlreg(list(m1,m2, m3, m4))| Model 1 | Model 2 | Model 3 | Model 4 | ||
|---|---|---|---|---|---|
| (Intercept) | 52.72*** | 56.43*** | 56.04*** | 53.02*** | |
| (1.98) | (1.14) | (11.15) | (1.28) | ||
| raceBlack | -16.50*** | -16.33** | -16.92*** | ||
| (0.59) | (5.36) | (1.04) | |||
| genderFemale | -13.16*** | -13.47 | -13.64*** | ||
| (0.42) | (14.76) | (1.00) | |||
| c_edu | 10.91*** | 10.85*** | 8.47*** | ||
| (0.09) | (0.09) | (0.41) | |||
| c_age | 1.27*** | 1.27*** | 1.26*** | ||
| (0.03) | (0.03) | (0.03) | |||
| c_hour | 1.67*** | 1.78*** | 1.64*** | ||
| (0.02) | (0.02) | (0.02) | |||
| raceBlack:c_edu | -2.42*** | -2.41*** | -3.22*** | ||
| (0.23) | (0.23) | (0.32) | |||
| genderFemale:c_edu | -4.77*** | -4.63*** | -2.39*** | ||
| (0.14) | (0.14) | (0.16) | |||
| raceBlack:c_hour | -0.70*** | -0.69*** | -0.79*** | ||
| (0.05) | (0.05) | (0.07) | |||
| genderFemale:c_hour | -0.26*** | -0.46*** | |||
| (0.04) | (0.04) | ||||
| raceBlack:genderFemale | 12.86*** | ||||
| (1.19) | |||||
| c_edu:c_hour | 0.44*** | ||||
| (0.01) | |||||
| raceBlack:genderFemale:c_edu | 1.35** | ||||
| (0.47) | |||||
| raceBlack:genderFemale:c_hour | 0.42*** | ||||
| (0.10) | |||||
| raceBlack:c_edu:c_hour | -0.16*** | ||||
| (0.03) | |||||
| genderFemale:c_edu:c_hour | -0.14*** | ||||
| (0.01) | |||||
| raceBlack:genderFemale:c_edu:c_hour | 0.10* | ||||
| (0.04) | |||||
| AIC | 1284753.11 | 1249271.03 | 1249227.47 | 1244209.93 | |
| BIC | 1284782.00 | 1249376.97 | 1249362.30 | 1244479.58 | |
| Log Likelihood | -642373.56 | -624624.52 | -624599.74 | -622076.96 | |
| Num. obs. | 112472 | 112472 | 112472 | 112472 | |
| Num. groups: STATEFIP | 42 | 42 | 42 | 42 | |
| Var: STATEFIP (Intercept) | 155.28 | 46.56 | 46.73 | 60.05 | |
| Var: Residual | 5343.69 | 3898.01 | 3896.16 | 3722.39 | |
| Num. groups: RACE | 2 | ||||
| Num. groups: SEX | 2 | ||||
| Var: RACE (Intercept) | 14.18 | ||||
| Var: SEX (Intercept) | 108.82 | ||||
| Var: STATEFIP genderFemale | 26.99 | ||||
| Var: STATEFIP c_edu | 6.21 | ||||
| Var: STATEFIP raceBlack | 17.47 | ||||
| Cov: STATEFIP (Intercept) genderFemale | -37.92 | ||||
| Cov: STATEFIP (Intercept) c_edu | 18.30 | ||||
| Cov: STATEFIP (Intercept) raceBlack | -31.92 | ||||
| Cov: STATEFIP genderFemale c_edu | -11.99 | ||||
| Cov: STATEFIP genderFemale raceBlack | 19.87 | ||||
| Cov: STATEFIP c_edu raceBlack | -10.21 | ||||
| p < 0.001, p < 0.01, p < 0.05 | |||||
library(merTools)
feSims <- FEsim(m4, n.sims = 100)plotFEsim(feSims) +
theme_bw() + labs(title = "Coefficient Plot of Model 4",
x = "Median Effect Estimate", y = "Total Income")
reSims <- REsim(m4, n.sims = 100)plotREsim(REsim(m4, n.sims = 100), stat = 'median', sd = TRUE)
library(doParallel)
library(foreach)
PI.time <- system.time(
PI <- predictInterval(merMod = m4, newdata = sub_income3,
level = 0.95, n.sims = 1000,
stat = "median", type="linear.prediction",
include.resid.var = TRUE))library(ggplot2);
ggplot(aes(x=1:50, y=fit, ymin=lwr, ymax=upr), data=PI[1:50,]) +
geom_point() +
geom_linerange() +
labs(x="Index", y="Prediction w/ 95% PI") + theme_bw()
Results
According to AIC of the above models, model 4 was considered as the best fitting models and was used for analyzing the results. This model included 5 explanatory variables(race, gender, education, age, working hours), and both random intercept and random slopes. In this model, the baseline group consisted of White males with age of 36 years who had 1 year of college education and who work 40 hours in a week. According to this model, estimated mean of the baseline group was 53.02 thousand. Income of Black males and White females were respectively 16.92 thousand and 13.64 thousand less compared to White males with baseline characteristics in terms of age and hours of work, but in case of Black females, the average income was 12.86 thousand higher than baseline groups. Effect of education on income was relatively less for Black males and White females, but relatively high for Black female compared to White males. For White males with baseline characteristics, every single year increase of educational attainment led 8.48 thousand increase in income, which was (8.48-3.22) =5.26 thousands for Black males and (8.46-2.39) = 6.09 thousand for White females with similar baseline characteristics in terms of age and working hours. On the other hand, for Black females, every single year increase of educational attainment led (8.48+1.35) =9.83 thousand increase in income. One hour increases of work in a week led 0.79 thousand and 0.46 thousand less income for Black male and White female respectively and 1.35 thousand more income compared to White male (1.64 thousand)/baseline group. Interaction effect of education and working hours was also found in this model.
So, according to the fixed effect component of the model, the effect of education mediated the relationships among race, gender, work hour and income. According to the random variance of the table, effect of gender, education, and race vary across States. The variation of sex and race was much greater than the variation of education.
Discussion
The purpose of this analysis was to understand the effect of education on the relationships among gender, race, and income as well as to understand whether the relationship varies across the States of USA. The result suggested that when holding for baseline age (36), education (1-year college education) and working hour (40hours/week) constant, estimated average income was higher for black females (65.87 thousand) than other groups (White male=53.02 thousand; White female=39.38 thousand; Black male=36.10). The results also suggested that education affect differently for males and females depending on their racial characteristics. Effect of education on income was relatively higher for Black females compared to Black males and Whites (both male and female). Income variation for every single hour of work was also revealed in terms of race and gender. The rise of income for the Black female due to the increase of every single hour was relatively higher than other counterparts.
The coefficient plot of model 4 was also revealed that the effect of education was higher for black female compared to white male, white female and black male.
Intra class correlation (ICC) of model 1(0.03) was very low which suggested that only 3% of the variation of income was accounted for by the States the people were from. By including people level variables (race, gender, education, working hour, age) in the model and the effect of gender, race and education to be varied by the States in model 4, the ICC this model was decreased (0.02) which suggested that still 2% of the unexplained variation can be accounted for by the State the people were from. The ICC could be further reduced by adding some State-level variables.
Limitation
The findings of the present study could only be generalized for the employed Black and Whites people with the age range of 18-66years and only for those States, the people were from. The State level variables were not included in this study. By adding State level variables might provide an in-depth understanding of the effect of education on income. In addition, the present analysis did not take into account for some other variables relevant to income such as occupational standing, or source of income (business, salaried work or wages).
Conclusion
The effect of education differs in terms of gender and race in determining income. The effect of education was foung in both gender gap of income and black-white gap of income as well as the income gap results from the combination of gender or race.
Reference
1.Campbell, L.; Kaufman, R. (2006).“Racial Differences in Household Wealth: Beyond Black and White”.Research in Social Stratification and Mobility.24(2): 131–52.
2.Couch, Kenneth; Daly, Mary C. (2002).“Black-White Wage Inequality in the 1990s: a Decade of Progress”(PDF).Economic Inquiry.40(1): 31–41.
3.Hegewisch, A. & Williams-Baron, E.(2018).“The Gender Wage Gap: 2017 Earnings Differences by Race and Ethnicity”. Pay Equality & Discrimination
4.Waters, Mary C.; Eschbach, Karl (1995).“Immigration and Ethnic and Racial Inequality in the United States”(PDF).Annual Review of Sociology.21(1): 419- 46.